[Bug target/94556] New: [10 Regression] FAIL: nptl/tst-thread-exit-clobber

[Bug target/94556] New: [10 Regression] FAIL: nptl/tst-thread-exit-clobber

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

            Bug ID: 94556
           Summary: [10 Regression] FAIL: nptl/tst-thread-exit-clobber
           Product: gcc
           Version: 10.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: hjl.tools at gmail dot com
  Target Milestone: ---
            Target: x86-64

On Linux/x32, GCC 10.0.1 20200324 caused

FAIL: nptl/tst-thread-exit-clobber

[hjl@gnu-cfl-2 build-x86_64-linux]$ nptl/tst-thread-exit-clobber --direct
info: unsigned int, direct pthread_exit call
info: double, direct pthread_exit call
error: tst-thread-exit-clobber.cc:119: not true: value ==
magic_values_double.v2
info: unsigned int, indirect pthread_exit call
info: double, indirect pthread_exit call
error: 1 test failures
[hjl@gnu-cfl-2 build-x86_64-linux]$ 

in tests on glibc master branch.  GCC 9.3 is OK.

[Bug target/94556] [10 Regression] FAIL: nptl/tst-thread-exit-clobber

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
                 CC|                            |ubizjak at gmail dot com
   Target Milestone|---                         |10.0
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2020-04-10

--- Comment #1 from H.J. Lu <hjl.tools at gmail dot com> ---
It is caused by r10-2846:

commit bc4aa158c9490e76573bee3eec90f893b7d0b1ae
Author: Uros Bizjak <ubizjak@gmail.com>
Date:   Wed Aug 28 17:09:51 2019 +0200

    * config/i386/i386-features.c
            (general_scalar_chain::compute_convert_gain):
            Correct cost for double-word shifts.
            (general_scalar_to_vector_candidate_p): Reject count operands
            greater or equal to mode bitsize.

    From-SVN: r274994

[Bug target/94556] [10 Regression] FAIL: nptl/tst-thread-exit-clobber

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

--- Comment #2 from H.J. Lu <hjl.tools at gmail dot com> ---
(In reply to H.J. Lu from comment #1)
> It is caused by r10-2846:
> 
> commit bc4aa158c9490e76573bee3eec90f893b7d0b1ae
> Author: Uros Bizjak <ubizjak@gmail.com>
> Date:   Wed Aug 28 17:09:51 2019 +0200
> 
>     * config/i386/i386-features.c
>             (general_scalar_chain::compute_convert_gain):
>             Correct cost for double-word shifts.
>             (general_scalar_to_vector_candidate_p): Reject count operands
>             greater or equal to mode bitsize.
>     
>     From-SVN: r274994

This isn't the real cause.  The bug is somewhere else.

[Bug target/94556] [10 Regression] FAIL: nptl/tst-thread-exit-clobber

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

--- Comment #3 from H.J. Lu <hjl.tools at gmail dot com> ---
Created attachment 48256
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=48256&action=edit
A tescase

[hjl@gnu-cfl-2 tmp]$ /usr/gcc-9.3.1-x32/bin/g++ -mx32 -O2 foo.cc -lpthread
[hjl@gnu-cfl-2 tmp]$ ./a.out 
info: double, direct pthread_exit call
[hjl@gnu-cfl-2 tmp]$ /usr/gcc-10.0.1-x32/bin/g++ -mx32 -O2 foo.cc -lpthread
[hjl@gnu-cfl-2 tmp]$ ./a.out 
info: double, direct pthread_exit call
Aborted (core dumped)
[hjl@gnu-cfl-2 tmp]$ 

CFI may be incorrect.

[Bug target/94556] [10 Regression] FAIL: nptl/tst-thread-exit-clobber

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Depends on|                            |83641

--- Comment #4 from H.J. Lu <hjl.tools at gmail dot com> ---
This is related to PR 83641.


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83641
[Bug 83641] -fstack-clash-protection generates incorrect CFI on i386

[Bug target/94556] [10 Regression] FAIL: nptl/tst-thread-exit-clobber

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

--- Comment #5 from H.J. Lu <hjl.tools at gmail dot com> ---
The code looks like:

        vmovq   %xmm2, %r15
        vmovq   %xmm3, %rbp
.LEHB0:
        call    pthread_exit
.LEHE0:
.L15:
        movl    %eax, %ebx
.L14:
        vmovq   %rbp, %xmm0
        movl    $4, %edi
        vzeroupper
        call    _Z11check_magicid

101             vmovq   %xmm3, %rbp
(gdb) next
103             call    pthread_exit
(gdb) p/x $rbp
$3 = 0x3f853f3c88fc5280
(gdb) next
106             movl    %eax, %ebx
(gdb) p/x $rbp
$4 = 0x88fc5280  <<<<< Only lower 32-bits are restored.
(gdb)

[Bug target/94556] [10 Regression] FAIL: nptl/tst-thread-exit-clobber

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |hjl.tools at gmail dot com

--- Comment #6 from H.J. Lu <hjl.tools at gmail dot com> ---
Created attachment 48258
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=48258&action=edit
A patch

I am testing this.

[Bug target/94556] [10 Regression] FAIL: nptl/tst-thread-exit-clobber

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

--- Comment #7 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by H.J. Lu <hjl@gcc.gnu.org>:

https://gcc.gnu.org/g:efc1f3577f38bb213b313661c025ac965baee953

commit r10-7697-gefc1f3577f38bb213b313661c025ac965baee953
Author: H.J. Lu <hjl.tools@gmail.com>
Date:   Mon Apr 13 04:23:05 2020 -0700

    x86: Restore the frame pointer in word_mode

    We must restore the frame pointer in word_mode for eh_return epilogues
    since the upper 32 bits of RBP register can have any values.

    Tested on Linux/x32 and Linux/x86-64.

            PR target/94556
            * config/i386/i386.c (ix86_expand_epilogue): Restore the frame
            pointer in word_mode for eh_return epilogues.

[Bug target/94556] FAIL: nptl/tst-thread-exit-clobber

[law at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |law at redhat dot com
            Summary|[10 Regression] FAIL:       |FAIL:
                   |nptl/tst-thread-exit-clobbe |nptl/tst-thread-exit-clobbe
                   |r                           |r

--- Comment #8 from Jeffrey A. Law <law at redhat dot com> ---
Fixed for gcc-10 AFAICT, it looks like it's latent on gcc-9, so leaving open in
case a backport is in the works.

[Bug target/94556] FAIL: nptl/tst-thread-exit-clobber

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |10.0
   Target Milestone|10.0                        |---

[Bug target/94556] FAIL: nptl/tst-thread-exit-clobber

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

--- Comment #9 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-9 branch has been updated by H.J. Lu <hjl@gcc.gnu.org>:

https://gcc.gnu.org/g:47ea8616d4f5fee875e0849e393575e00def5894

commit r9-8517-g47ea8616d4f5fee875e0849e393575e00def5894
Author: H.J. Lu <hjl.tools@gmail.com>
Date:   Mon Apr 20 05:51:29 2020 -0700

    x86: Restore the frame pointer in word_mode

    We must restore the frame pointer in word_mode for eh_return epilogues
    since the upper 32 bits of RBP register can have any values.

    Tested on Linux/x32 and Linux/x86-64.

            Backport from master
            PR target/94556
            * config/i386/i386.c (ix86_expand_epilogue): Restore the frame
            pointer in word_mode for eh_return epilogues.

    (cherry picked from commit efc1f3577f38bb213b313661c025ac965baee953)

[Bug target/94556] FAIL: nptl/tst-thread-exit-clobber

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

--- Comment #10 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-8 branch has been updated by H.J. Lu <hjl@gcc.gnu.org>:

https://gcc.gnu.org/g:472c004fcd13e429dad7d1b829d21e26d24e39a4

commit r8-10192-g472c004fcd13e429dad7d1b829d21e26d24e39a4
Author: H.J. Lu <hjl.tools@gmail.com>
Date:   Mon Apr 20 05:51:29 2020 -0700

    x86: Restore the frame pointer in word_mode

    We must restore the frame pointer in word_mode for eh_return epilogues
    since the upper 32 bits of RBP register can have any values.

    Tested on Linux/x32 and Linux/x86-64.

            Backport from master
            PR target/94556
            * config/i386/i386.c (ix86_expand_epilogue): Restore the frame
            pointer in word_mode for eh_return epilogues.

    (cherry picked from commit efc1f3577f38bb213b313661c025ac965baee953)

[Bug target/94556] FAIL: nptl/tst-thread-exit-clobber

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94556

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED
   Target Milestone|---                         |8.5

--- Comment #11 from H.J. Lu <hjl.tools at gmail dot com> ---
Fixed for GCC 10, GCC 9.4 and GCC 8.5.